<h2 id="recommended-settings">📋 Recommended Settings / 推荐设置</h2><ul><li>Base Model / 底模: * Z-Image (ZIT) Base</li><li>LoRA Weight / 推荐权重: * Standard: 1.0<ul><li>For Turbo Models: 1.2 - 1.4 (Turbo 模型建议适当调高权重以获得更好的还原度)</li><li>CFG Scale / 引导值: * 4</li><li>Sampling Steps / 采样步数: * 30 - 50</li></ul></li></ul><hr /><h2 id="model-description">📝 Model Description / 模型介绍</h2>[EN] This LoRA is meticulously trained based on the Z-Image (ZIT) Base model. <ul><li>Weight Advice: While 1.0 works perfectly for most scenarios, please increase the weight if you are using Turbo-distilled models to maintain character consistency.</li><li>Prompting: Works best with natural language descriptions. Ensure CFG stays around 4 to avoid over-saturation.</li></ul>[CN] 本模型基于 Z-Image (ZIT) 底模进行精细训练。<ul><li>权重建议： 默认权重 1.0 在大多数场景下表现完美。如果你使用的是 Turbo 系列的高速模型，建议将权重提升至 1.2 - 1.4 以确保角色的一致性。</li><li>提示词建议： 推荐使用自然语言（句子）进行描述。请保持 CFG 在 4 左右，过高可能会导致画面崩坏或色彩过饱和。</li></ul>

Files with SHA256: 30b8f061464fd8e99c1215b956d1b0cdf5fc21287eedc520d3bce28166f8c3dc